emily2040/seedance-2.0

Seedance 2.0 Skill OS is a package of instructions and reference material that teaches an AI assistant how to direct AI-generated video using a video generation platform called Seedance 2.0. Rather than asking the AI model for a generic cinematic look, this project encodes the way a real film director thinks: figuring out what a scene is actually doing dramatically, then making the camera angle, lighting, character movement, performance, and sound all serve that one idea, consistently across an entire story. The repository is organized into many small skill files, each covering a specific part of the filmmaking workflow. Together they guide the assistant on when to ask the user clarifying questions before writing a prompt, how to write both detailed and compact prompts for different generation modes such as text to video, image to video, video to video, and first and last frame workflows, and how to keep track of which reference images or clips represent a character, a location, a camera move, or an audio cue. It also includes guidance for rewriting requests that might involve copyrighted characters into safer original versions, and for troubleshooting a generation that came out wrong by trying one change at a time rather than guessing randomly. The project supports multiple languages beyond English, including Chinese, Japanese, Korean, Spanish, and Russian, each with their own quickstart guide and vocabulary reference so users can write prompts naturally in their own language. The repository ships with dozens of fully worked examples covering different genres, from product ads and music videos to horror, anime, action, comedy, and documentary style. It also keeps notes on the specific platforms where Seedance 2.0 is available, dated to when those details were verified, rather than guessing at pricing or model names. The project is released under the MIT license. The full README is longer than what was shown.